When I first heard about Andy Rubin announcing Android will be optimized for Intel Atom chips for all versions of Android from now on, I was very disappointed with Google, because I thought it was a poor strategic decision on their part. It felt backwards. Even Microsoft is moving to ARM, and I bet Apple will make a Transformer-like clamshell or at least a Macbook Air-like one based on ARM, soon, too. So why would Google support “the past”, instead of looking forward ?
ARM chips keep advancing fast, at about 2.5x improvement each year, or about 4x every 18 months. That’s twice as fast as Moore’s Law for x86 chips. And best of all, they advance that much, while still keeping the same overall low power consumption. And I bet that’s all chip makers’ goal – to improve their chip as much as they can, every year, while maintaining the same power consumption.
What this means is that ARM chips are closing the performance gap on Intel/AMD chips fast, and chips like Kal-El should already feel like they are more than enough performance wise, for most people, whether it’s on a phone, or even on a tablet. With a Kal-El chip, page rendering of complex and “full” web pages should happen in about half a second, and it will feel almost instant. Games will look great, too, and it’s one of the reasons I’ve been wanting Google to transform Google TV in a “console platform”, with set-top boxes based on Kal-El or other such powerful quad core chips.
Plus, the ARM competition is fierce compared to the x86 market, where you only have 2 competitors: Intel and AMD. In the ARM market, you have the 4 main ones: Nvidia, Qualcomm, TI, and of course Samsung. But then you also have Freescale (their quad core 1 Ghz Cortex A9 might be in PS Vita), ST-Ericsson, Marvell, and soon LG (they licensed Cortex A15 from ARM recently).
That’s 8 competitors! What more could Google want in terms of chip competition? Why would they still need Intel to provide competition, when they’re still stumbling and crashing in the mobile market with their Atom CPU?
My first guess was that this decision was mostly political. I mean, Paul Otellini, Intel’s CEO, is on Google’s board for crying out loud. Then they have strong relationships in the server market, since Google build their own servers, and obviously they have strong ties with Intel there, since they’re probably their biggest customer.
They’ve also partnered on Google TV, which I still think was a huge mistake on Google’s part. It’s the main reason why Google TV failed. Well, Logitech’s set top box costing $300 was the main reason, but Intel had a big part in that. Their Atom CE, should cost somewhere between $75-$100, as a component. That translates to about $150-$200 from the retail price. If you’ve been watching iFixit’s disassembling of phones and tablets, you’ll see the components end up about half the retail price. Meanwhile, a Tegra 2 costs about $20, so with it, the Logitech Revue would cost about $150 at most.
Then, Google partnered with Intel again on Chromebooks – and again the pricing seemed significantly more than people expected. Granted, people were comparing the dual core 1.8 Ghz Atom with a single core older 1.6 Ghz Atom in cheaper netbooks, but I believe they were ultimately right. The Chromebooks are too expensive for what they are supposed to do – just browse the web. It’s irrelevant if it even comes with a quad core i7 in them. People are not going to pay hundreds more for that, when it’s still only for browsing the web.
That’s the main mistake Motorola and HTC did with their tablets, too. The $500 price is about the maximum people are willing to pay for such a consumption device, regardless of how many “specs” you throw in there.
So why didn’t Google go with ARM, and make them $200-$250? That would’ve been the sweet-spot for Chromebooks, instead of $350-$500. They’ve said they will, so hopefully we’ll see some $250 Kal-El based Chromebooks, soon. But it was still a mistake to start out like that, and get negative reviews on it as first impressions. They might affect their growth in the future.
I’m really hoping this is not another one of such mistakes by partnering again with Intel, because if Google actually helps Intel get a hold of the mobile market, they’ll be hurting themselves rather than helping themselves in the end. Here’s why.
Why does Microsoft go to ARM now? They are very vulnerable there, starting from scratch with no apps. They are much behind even Android on tablets with Windows 8 for ARM. Android already has at least hundreds of tablet optimized apps, and they already have the 200,000 or so mobile apps, where most can still be scaled up pretty nicely on a tablet.
The reason is because Microsoft thinks Intel can’t pull it off. Not only do they think Intel won’t achieve the kind of performance/power consumption/price that ARM chips have (or they’ll do it too late), but they also think they won’t be well received by the market, even if they do. It wouldn’t be the first time when even a big player from another market, was rejected by a new market. So they don’t want to bet their farm on Intel.
But here’s the thing. Microsoft would love for Intel to become a strong player in the mobile market, because then they can start promoting Intel-based tablets again, where their real strength lies (with x86 legacy programs). So if Google actually helps Intel become a strong player in the mobile market through Android, Google would barely benefit from it, since like I said, chip competition is very strong already in the mobile market, but Microsoft would benefit a lot more.
So is there more to this than just a political decision meant to help Intel, but with little benefit to Google? I believe there may be and hopefully I’m right, because I would hate for Google to make such a bad strategic decision, otherwise.
Here’s what I think is the real reason why Google may be optimizing Android for Intel’s chips: Google wants to become your main OS, not just on smartphones and tablets, but also on all older and newer x86 systems.
If that’s the real reason, then it may not be such a bad decision after all, and in fact it’s pretty smart. As Android becomes a more mature and even more popular OS, you will be able to do more and more with it, and once you get more professional applications, you’ll want as much performance as possible, too. It will take a while until ARM reaches Intel’s most powerful chips (even though it seems Intel has all but stopped improving the performance of the Ivy Bridge CPU, and they want to focus more on lowering power consumption), and Google may not want to wait until that happens.
Plus, with so many hundreds of millions, if not billions of PC’s out there in the world, Google may want to make it easy for Android to be installed on those machines one day soon. Of course, it’s not just Intel chips they’d have to worry about if they want to make that happen, but it would be one huge obstacle cleared out of the way. Now that Intel has Google’s ear, AMD will probably ask Google to optimize Android for AMD chips, as well, and in return, Google may also ask them to support Android on other older machines of theirs.
So ideally, Google would get both Intel and AMD to make drivers for Android for their other older chips, while “pretending” to help them get a hold of the mobile market, with no significant outcome that would favor Intel or AMD. That would be the best case scenario for Google, without compromising themselves much in the mobile market, and without leaving the door open for Microsoft.
What do you guys think? Did Google do the right thing partnering with Intel on this, or should they have stayed exclusively with ARM and let Intel fade away in the “PC world” and hurting Microsoft in the same time, while they move on with the “Post-PC world”?
Like this post? Share it!
It’s about time google entered the computer OS market. Well I hope they are…
how about google wanting to use android on more powerful hardware to build sagetv media servers to provide content to less powerful google tv devices and phones/tablets ..
What a delusional idiot.
Atom aside, which is not even intel’s main focus, no one can make anything nearly as fast as intel from notebook lv to desktop lv, if you want an OS on EVERYTHING, Intel is the only choice now. Arm is not getting there anytime soon, Kal-El will have a hard time catch up to atom which itself is too weak for a full featured notebook.
Let me tell you something, it’s easier to build a simple thing when you already mastered building something far more complex, Intel shows that with Ivy bridge, not only they are the best, but they outclass everyone by a large margin. On other hand Arm has to design something that is far more complex than what they have, work with firms that is 12-18 month behind Intel in Fab process, that include every other firm on earth, yes get it right, Intel is THAT good, they are on 32nm for almost 2 years and is moving to 22nm, while everyone else is still on 40-45 and looking to move to 28nm.(their 28nm will likely ship behind Intel’s 22nm…)
Intel is best processor maker on earth, and it will not change in foreseeable future.
Of course Intel’s Core i chips are much more powerful than ARM, but that might be irrelevant for most people who want a personal computing device, because for all intents an purposes a quad core ARM chip will appear to be just as fast as Intel’s chips to most people, unless they want to do heavy specialized work on it, but there aren’t any apps for that anyway, right now.
And I disagree with your first point that it’s much easier to make something simpler. Intel has promised a low power chip for smartphones since 2007 or so, and we still haven’t seen it yet, and a true low power chip that consumes just as much power as an ARM chip might not even be available until at least 2013. And who knows what kind of performance will have. For example, a single core Atom chip uses like 5W of power now, but a dual core Atom uses 10W of power. So Intel has to catch-up to ARM in power consumption, but they also need to keep advancing the performance of ARM so ARM doesn’t surpass Atom, and that may have already happened with Kal-El. So, my point is, it’s not easy at all for Intel.
Oh, and your argument could stand on its own without name calling, too ;). But, thanks for the comment, anyway.
Great article. Its very possible that Google wants android to work as a desktop OS as well. But, if that were the case, why would Google develop its ChromeOS still? It seems to me that the issue here is a difference in the architecture of the x86 Intel chips and the ARM chips. By architectural design the x86 use more power but have more horse power. ARM is designed for less power consumption and less power usage. Its like trying to modify a 4 cylinder to tow as good as a v8 or make a v8 have better gas mileage as a 4 cylinder. Yes, Intel has the most powerful x86 chips on the market but they were not designed from the ground up to be power efficient. Im not saying that it cannot be done, and if anyone could do it they would be Intel.
I would love to see the Android OS mature to the point of running on desktop computers. I worry about how much desktops are in the future though. It seems most everything is moving to the tablet future. I see the public moving away from desktops all together.
Either way, I think Google knows what they are doing.
Give me a break, first of there are dual core atom has TDP lower than 8W, second, I won’t go into details but power usage and TDP are not the same thing. Most importantly, all of them are on outdated 45nm process because they are made for netbooks not phones, second they are not intel’s focus, so they are more or less like red haired step child of Intel. 32nm are already a very different story, 22nm chips will be even better.
Also phones aside, For tablets the screen is using so much power anyway, 5-10W difference when your Chip is running at MAX won’t be that big a difference, I can definitely see future high performance tablet running on something like the process they talked about in this article:
this is the thing with x86. The ISA was rushed to market and patched over and over and over. The reason it has sustained market share was because it was the first to market. The CISC heavy ISA of x86 finally being caught up by the RISC oriented ARM. CISC simply can’t beat RISC at power consumption. It’s only a matter of time when the fab process catches up.
This move only makes sense if Google is building an OS for laptops and desktops as the article says. Assuming that Google does build an OS for laptops and desktops then this makes sense. They should just leave ARM chips in the Android phones/tablets and use the Intel chips only for the assumed laptop/desktop OS.
Great move by Google..but ARM processors are not common for PCs and for people who love RAW power always consider Intel, thus making Android support Intel chips would screw away Windows…may be doomsday for closed source is not far away..!!!